[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Why rewrite the Subversion working copy?

From: <kmradke_at_rockwellcollins.com>
Date: Wed, 23 Apr 2008 11:49:29 -0500

Listman <listman_at_burble.net> wrote on 04/23/2008 11:31:50 AM:
> On Apr 23, 2008, at 6:05 AM- Apr 23, 2008, Bert Huijben wrote:
> >
> > Not sure if everybody noticed this. The original questions on the p4
> > edit
> > speed comparison all used working copies on 'netapp's.. (That is: A
> > remote
> > directory on a NAS), while we optimize our working copy mostly for
> > local
> > filesystems..
> >
> > I don't think any other scanning solutions could fix performance on
> > those
> > systems (Unless we trigger something that makes the network caching
> > fail).
> > That network layout almost requires a checkin-checkout solution
> > which we
> > don't.
> > (I'm not sure if we should even spend time to speed up that subversion
> > usage)
> >
> but thats the reality of most corporations, user accounts (and hence
> working copies) are on a remote dir, on a NAS. Perforce performs well
> in this situation because the "p4 edit" command negates the need for
> disk-scanning the wc. I really think we should have this as an
> optional mode with SVN.

APR isn't the most efficient at some of these operations either. For
example, I have an open bug report to increase the default copy
buffer size from 512 bytes to something larger. Seeing a network
trace of a svn working copy stored on the network can be a real
eye opener... (We were seeing millions of partially filled packets
that were waiting for acks before the next one was sent...)

Kevin R.

To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-04-23 18:50:03 CEST

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.